Implementation on Automatic Number Plate Recognition in a video| TensorFlow| Explanation with code|

Our Take

Credit to RISAi for putting a technical face on something we usually only see in polished marketing decks. This tutorial isn't about Flock Safety specifically, but it's a useful reality check: the core computer-vision pipeline behind automated license plate recognition — detect plate, crop, run OCR, output text — isn't some proprietary black box. It's a weekend project with open-source libraries and a training set. Companies like Flock aren't selling magic; they're selling scale, hosting, and a searchable nationwide network built on top of tech like this.

That distinction matters. The debate over ALPR surveillance often gets stuck on 'how sophisticated is the AI,' when the real issue is what happens after a plate gets read: who stores it, how long, who can query it, and whether your daily commute becomes a permanent, cross-jurisdictional record. A tutorial like this shows the recognition step is commoditized. The surveillance risk lives entirely in the network effect — thousands of cameras, one database, zero warrant required in most places.

If you want to see that network effect in your own neighborhood, check our camera map to find documented Flock installations near you, and visit take action for concrete steps to push back on unchecked ALPR deployment in your city or county.